Full Coverage Submetering of Diverse Electrical Loads
Understanding consumer electricity use requires measuring the power draw of both major appliances and the long tail of the miscellaneous electrical loads that comprise a growing fraction of the total in residential and commercial settings. This paper revisits the assumption that full coverage of household loads is impractical by implementing a system for direct device submetering on plugloads and overhead lighting, and deploying it in seven homes and one office. In doing so, we generate a dataset that demonstrates the growing importance of these loads, with 51% of total electrical energy consumed by plug-loads and lighting. Additionally, we find that the long tail of consumption cannot be discounted. In this dataset, devices that draw less than 30 W account for 30% of the submetered energy use and 16% of the total. To generate this dataset, we deploy 366 nearly invisible meters across these locations. Supporting such small and densely deployed meters requires rethinking previously held assumptions about the network architecture and the need for a mesh. We utilize Bluetooth Low Energy and explore its performance in a star topology for data collection. Our modest-scale deployment demonstrates that submetering with coverage, longevity, and unobtrusiveness is indeed viable, enabling new observations and insights into increasingly scattered electrical consumption.